scispace - formally typeset
Search or ask a question
Institution

Facebook

CompanyTel Aviv, Israel
About: Facebook is a company organization based out in Tel Aviv, Israel. It is known for research contribution in the topics: Artificial neural network & Language model. The organization has 7856 authors who have published 10906 publications receiving 570123 citations. The organization is also known as: facebook.com & FB.


Papers
More filters
Proceedings ArticleDOI
06 Jun 2010
TL;DR: This paper presents how Scribe, Hadoop and Hive together form the cornerstones of the log collection, storage and analytics infrastructure at Facebook and enabled us to implement a data warehouse that stores more than 15PB of data and loads more than 60TB of new data every day.
Abstract: Scalable analysis on large data sets has been core to the functions of a number of teams at Facebook - both engineering and non-engineering. Apart from ad hoc analysis of data and creation of business intelligence dashboards by analysts across the company, a number of Facebook's site features are also based on analyzing large data sets. These features range from simple reporting applications like Insights for the Facebook Advertisers, to more advanced kinds such as friend recommendations. In order to support this diversity of use cases on the ever increasing amount of data, a flexible infrastructure that scales up in a cost effective manner, is critical. We have leveraged, authored and contributed to a number of open source technologies in order to address these requirements at Facebook. These include Scribe, Hadoop and Hive which together form the cornerstones of the log collection, storage and analytics infrastructure at Facebook. In this paper we will present how these systems have come together and enabled us to implement a data warehouse that stores more than 15PB of data (2.5PB after compression) and loads more than 60TB of new data (10TB after compression) every day. We discuss the motivations behind our design choices, the capabilities of this solution, the challenges that we face in day today operations and future capabilities and improvements that we are working on.

455 citations

Journal ArticleDOI
03 Mar 2020-JAMA
TL;DR: National spending estimates stratified by health condition, age group, sex, type of care, and type of payer and modeled for each year from 1996 through 2016 show low back and neck pain had the highest amount of health care spending in 2016.
Abstract: Importance US health care spending has continued to increase and now accounts for 18% of the US economy, although little is known about how spending on each health condition varies by payer, and how these amounts have changed over time. Objective To estimate US spending on health care according to 3 types of payers (public insurance [including Medicare, Medicaid, and other government programs], private insurance, or out-of-pocket payments) and by health condition, age group, sex, and type of care for 1996 through 2016. Design and Setting Government budgets, insurance claims, facility records, household surveys, and official US records from 1996 through 2016 were collected to estimate spending for 154 health conditions. Spending growth rates (standardized by population size and age group) were calculated for each type of payer and health condition. Exposures Ambulatory care, inpatient care, nursing care facility stay, emergency department care, dental care, and purchase of prescribed pharmaceuticals in a retail setting. Main Outcomes and Measures National spending estimates stratified by health condition, age group, sex, type of care, and type of payer and modeled for each year from 1996 through 2016. Results Total health care spending increased from an estimated $1.4 trillion in 1996 (13.3% of gross domestic product [GDP]; $5259 per person) to an estimated $3.1 trillion in 2016 (17.9% of GDP; $9655 per person); 85.2% of that spending was included in this study. In 2016, an estimated 48.0% (95% CI, 48.0%-48.0%) of health care spending was paid by private insurance, 42.6% (95% CI, 42.5%-42.6%) by public insurance, and 9.4% (95% CI, 9.4%-9.4%) by out-of-pocket payments. In 2016, among the 154 conditions, low back and neck pain had the highest amount of health care spending with an estimated $134.5 billion (95% CI, $122.4-$146.9 billion) in spending, of which 57.2% (95% CI, 52.2%-61.2%) was paid by private insurance, 33.7% (95% CI, 30.0%-38.4%) by public insurance, and 9.2% (95% CI, 8.3%-10.4%) by out-of-pocket payments. Other musculoskeletal disorders accounted for the second highest amount of health care spending (estimated at $129.8 billion [95% CI, $116.3-$149.7 billion]) and most had private insurance (56.4% [95% CI, 52.6%-59.3%]). Diabetes accounted for the third highest amount of the health care spending (estimated at $111.2 billion [95% CI, $105.7-$115.9 billion]) and most had public insurance (49.8% [95% CI, 44.4%-56.0%]). Other conditions estimated to have substantial health care spending in 2016 were ischemic heart disease ($89.3 billion [95% CI, $81.1-$95.5 billion]), falls ($87.4 billion [95% CI, $75.0-$100.1 billion]), urinary diseases ($86.0 billion [95% CI, $76.3-$95.9 billion]), skin and subcutaneous diseases ($85.0 billion [95% CI, $80.5-$90.2 billion]), osteoarthritis ($80.0 billion [95% CI, $72.2-$86.1 billion]), dementias ($79.2 billion [95% CI, $67.6-$90.8 billion]), and hypertension ($79.0 billion [95% CI, $72.6-$86.8 billion]). The conditions with the highest spending varied by type of payer, age, sex, type of care, and year. After adjusting for changes in inflation, population size, and age groups, public insurance spending was estimated to have increased at an annualized rate of 2.9% (95% CI, 2.9%-2.9%); private insurance, 2.6% (95% CI, 2.6%-2.6%); and out-of-pocket payments, 1.1% (95% CI, 1.0%-1.1%). Conclusions and Relevance Estimates of US spending on health care showed substantial increases from 1996 through 2016, with the highest increases in population-adjusted spending by public insurance. Although spending on low back and neck pain, other musculoskeletal disorders, and diabetes accounted for the highest amounts of spending, the payers and the rates of change in annual spending growth rates varied considerably.

450 citations

Posted Content
TL;DR: Submanifold sparse convolutional networks (SSCNs) as discussed by the authors were proposed to process spatially-sparse data more efficiently, and use them to develop spatiallysparse CNNs.
Abstract: Convolutional networks are the de-facto standard for analyzing spatio-temporal data such as images, videos, and 3D shapes. Whilst some of this data is naturally dense (e.g., photos), many other data sources are inherently sparse. Examples include 3D point clouds that were obtained using a LiDAR scanner or RGB-D camera. Standard "dense" implementations of convolutional networks are very inefficient when applied on such sparse data. We introduce new sparse convolutional operations that are designed to process spatially-sparse data more efficiently, and use them to develop spatially-sparse convolutional networks. We demonstrate the strong performance of the resulting models, called submanifold sparse convolutional networks (SSCNs), on two tasks involving semantic segmentation of 3D point clouds. In particular, our models outperform all prior state-of-the-art on the test set of a recent semantic segmentation competition.

443 citations

Journal ArticleDOI
TL;DR: The authors of as discussed by the authors examined the nature of the reach and frequency of branded content on Facebook, and found that the reach of brand content among Friends of Fans significantly exceeded the reach among Fans, indicating that the value of a Fan can be assessed in three primary ways: increasing the depth of engagement and loyalty among Fans and generating incremental purchase behavior.
Abstract: The following white paper is based on a collaboration between comScore and Facebook. The research, which examines the nature of the reach and frequency of branded content on Facebook, is largely based on May 2011 findings from comScore Social Essentials™, but also includes insights from Facebook’s internal analytics platform. The summary below establishes the key findings and implications of this research. Facebook is the dominant social networking site with an audience of approximately 160 million U.S. visitors each month and accounts for 90 percent of all time spent on social networking sites. (Source: comScore) Branded content on social media can take many forms – content shared directly from brands, re-shared content from connections, or social marketing such as Facebook’s Sponsored Stories ad unit. While there has been tremendous focus on counting the incidence of these brand mentions on social networks (and, to a lesser degree, categorizing it), we find that the reach and frequency of social media brand impressions is far more important than simple counting statistics. Facebook users spend more than a quarter of their time on the site consuming and interacting with the Facebook Newsfeed, and this activity represents 4 percent of all time spent online in the U.S. The Newsfeed is also the primary location where branded content is consumed. In fact, users are 40-150 times more likely to consume branded content in the Newsfeed than to visit the Fan Page itself. (Source: comScore) There are typically two potential audiences for branded content on Facebook. Fans of brands on Facebook (those who have explicitly “liked” a brand) are the easiest to reach with social media brand impressions, but the Friends of those Fans also constitute an important incremental audience. Friends of Fans typically represent a much larger set of consumers (34 times larger, on average, for the top 100 brand pages) and can receive social media brand impressions by way of their Friends. In examples from this paper, the reach of branded content among Friends of Fans significantly exceeded the reach among Fans. When a brand focuses on acquiring and engaging Fans it can benefit from a significant secondary effect – exposure among Friends of Fans that often surpasses reach among Fans. In the case of some brands, Facebook Fans may have different aggregate demographic and behavioral profiles than typical brand purchasers, indicating that social media may require different approaches to marketing strategy. The “Value of a Fan” can be assessed in three primary ways: increasing the depth of engagement and loyalty among Fans, generating incremental purchase behavior, and leveraging the ability to influence Friends of Fans.

442 citations

Posted Content
TL;DR: This work proposes VoteNet, an end-to-end 3D object detection network based on a synergy of deep point set networks and Hough voting that achieves state-of-the-art 3D detection on two large datasets of real 3D scans, ScanNet and SUN RGB-D with a simple design, compact model size and high efficiency.
Abstract: Current 3D object detection methods are heavily influenced by 2D detectors. In order to leverage architectures in 2D detectors, they often convert 3D point clouds to regular grids (i.e., to voxel grids or to bird's eye view images), or rely on detection in 2D images to propose 3D boxes. Few works have attempted to directly detect objects in point clouds. In this work, we return to first principles to construct a 3D detection pipeline for point cloud data and as generic as possible. However, due to the sparse nature of the data -- samples from 2D manifolds in 3D space -- we face a major challenge when directly predicting bounding box parameters from scene points: a 3D object centroid can be far from any surface point thus hard to regress accurately in one step. To address the challenge, we propose VoteNet, an end-to-end 3D object detection network based on a synergy of deep point set networks and Hough voting. Our model achieves state-of-the-art 3D detection on two large datasets of real 3D scans, ScanNet and SUN RGB-D with a simple design, compact model size and high efficiency. Remarkably, VoteNet outperforms previous methods by using purely geometric information without relying on color images.

442 citations


Authors

Showing all 7875 results

NameH-indexPapersCitations
Yoshua Bengio2021033420313
Xiang Zhang1541733117576
Jitendra Malik151493165087
Trevor Darrell148678181113
Christopher D. Manning138499147595
Robert W. Heath128104973171
Pieter Abbeel12658970911
Yann LeCun121369171211
Li Fei-Fei120420145574
Jon Kleinberg11744487865
Sergey Levine11565259769
Richard Szeliski11335972019
Sanjeev Kumar113132554386
Bruce Neal10856187213
Larry S. Davis10769349714
Network Information
Related Institutions (5)
Google
39.8K papers, 2.1M citations

98% related

Microsoft
86.9K papers, 4.1M citations

96% related

Adobe Systems
8K papers, 214.7K citations

94% related

Carnegie Mellon University
104.3K papers, 5.9M citations

91% related

Performance
Metrics
No. of papers from the Institution in previous years
YearPapers
20241
202237
20211,738
20202,017
20191,607
20181,229